Yiran Tao

Hi! I am a PhD student at MIT, advised by Prof. Andreea Bobu.

My research focuses on robot learning and robotic manipulation, particularly Vision-Language-Action (VLA) models and human-in-the-loop learning for robotic systems.

Previously, I received my MS in Robotics from Carnegie Mellon University, where I worked with Prof. Zackory Erickson on shared autonomy and assistive robotics.

I obtained my bachelor's degree from Wuhan University, where I worked with Prof. Zhenzhong Chen on computer vision research.

Email  /  CV (Last updated: Mar. 2026)  /  Google Scholar  /  LinkedIn  /  Github

My current research focuses on robot manipulation and learning with Vision-Language-Action (VLA) models. I am interested in improving the robustness and controllability of VLA policies for real-world robotic tasks. My current work explores how VLA policies represent and execute multiple valid manipulation strategies in real-world tasks. I am also interested in how modifying robot observations before they are processed by VLA policies—such as removing distractors or simplifying scenes—can improve policy reliability without additional fine-tuning.

An early prototype experiment (developed as part of a course project) can be seen here: Video

We introduce LLM-Driven Automatic Mode Switching (LAMS), a method that automatically switches robot control modes based on task context using large language models.

We introduce ILSA, an Incrementally Learned Shared Autonomy framework that continually improves its assistive control policy through repeated user interactions.

Last updated: Mar. 2026   |   Template from Jon Barron